Memory Chip Executives Declare End of Classic Boom-Bust Cycle, Citing AI Demand
Key Facts
- Executives from Micron and Sandisk say the traditional boom-bust cycle in the memory industry has ended due to sustained AI-driven demand.
- Micron shares have skyrocketed 370% over the past year, while Sandisk has surged more than 1,100%.
- High-bandwidth memory (HBM) is reported sold out through 2026 with a projected $100 billion total addressable market by 2028.
- AI is viewed as an enduring force that smooths out the sector’s historic volatility, unlike previous cycles tied to consumer electronics and PCs.
- New semiconductor fabs cost $15 billion or more and take 18 months or longer to build, contributing to structural supply constraints.
The memory chip industry’s decades-long pattern of sharp booms followed by painful busts appears to be breaking, according to executives at leading suppliers Micron and Sandisk. Surging and sustained demand for specialized memory used in artificial intelligence systems is fundamentally altering the economics of the sector, they say.
This shift comes as the broader AI boom continues to reshape semiconductor supply chains. Unlike past cycles driven by personal computers, smartphones and consumer gadgets — which led to periodic overcapacity and price collapses — demand tied to AI training and inference is proving more predictable and longer-lasting.
AI Reshapes Memory Market Dynamics
Memory chips, particularly high-bandwidth memory (HBM) and other advanced DRAM variants, have become critical components in the data centers powering large AI models. These specialized chips enable the massive parallel processing required for training and running today’s most advanced AI systems.
According to industry reporting, HBM is already sold out through 2026, pointing to a multi-year shortage that contrasts sharply with historical patterns. Analysts project the total addressable market for AI-related memory could reach $100 billion by 2028, creating a structural tailwind that executives believe will prevent the usual bust phase.
The origins of the current tightness trace back to supply chain disruptions during the COVID-19 pandemic, when chip shortages prompted conservative capacity planning. Building new fabrication facilities now costs $15 billion or more and requires at least 18 months to complete, making suppliers extremely reluctant to over-expand. This lag virtually guarantees that new supply arrives well after demand peaks, traditionally flooding the market and crashing prices.
However, AI’s once-in-a-generation nature appears to be changing the calculus. Executives at Micron and Sandisk have publicly stated that the memory stock cycle of “boom-bust-repeat” is over, with AI providing a more stable and growing baseline of demand.
Stock Performance Reflects New Reality
Financial markets have responded enthusiastically to the changed outlook. Micron shares have risen 370% over the past year, while Sandisk has delivered even more dramatic returns, climbing more than 1,100%, according to CNBC reporting.
These gains stand in contrast to previous cycles when investors would rotate out of memory stocks at the first sign of inventory buildup or softening consumer demand. The current rally reflects a belief that AI spending will remain elevated for years, supported by major cloud providers and technology companies racing to build out AI infrastructure.
Bloomberg has reported that investors are “tearing up the memory stock playbook” as the AI boom redefines valuation metrics for the sector. Rather than trading purely on cyclical swings, memory companies are increasingly being valued on their exposure to long-term AI infrastructure buildouts.
Competitive Landscape and Industry Context
The memory sector has long been dominated by a handful of players, with Micron, Samsung and SK Hynix controlling the majority of DRAM and NAND flash production. Sandisk, a Western Digital brand focused on NAND flash and storage solutions, has also benefited from increased demand for AI-optimized storage systems.
Industry expert Thomas Coughlin, president of Coughlin Associates, has noted that the enormous capital requirements for new fabs — combined with long construction timelines — create natural constraints on supply growth. This structural barrier, once a source of volatility, may now act as a stabilizing force given AI’s persistent appetite for memory.
Other reports highlight that while current spending on AI infrastructure is high, the technology’s fundamental importance suggests it differs from previous hype cycles. Unlike the late 1990s dot-com era, today’s AI investments appear more measured, with companies demanding tangible results and performance improvements rather than speculative bets.
Impact on Developers, Users and the Industry
For AI developers and cloud providers, the ongoing memory shortage creates both challenges and opportunities. The HBM sell-out through 2026 means securing supply has become a strategic imperative, potentially favoring larger players with established relationships with memory manufacturers.
This dynamic could slow the pace of AI infrastructure expansion for some organizations while accelerating it for those able to lock in capacity. End users may ultimately see these constraints reflected in higher costs for AI services or longer wait times for access to the most advanced models.
Within the semiconductor industry, the shift is prompting a reevaluation of investment strategies. Companies are likely to focus more heavily on advanced packaging technologies and specialized memory solutions optimized for AI workloads rather than chasing broad commodity production.
The changed cycle also has implications for equipment suppliers, materials providers and the entire semiconductor ecosystem. A more stable demand environment could lead to more predictable capital expenditure patterns and potentially smoother growth across the supply chain.
What’s Next
Looking ahead, the industry will closely monitor whether AI demand continues to outpace supply through 2026 and beyond. The projected $100 billion market for AI memory by 2028 suggests significant room for growth, but execution risks remain around new fab construction and technological transitions to next-generation memory solutions.
Micron and Sandisk are expected to provide further updates on their capacity expansion plans and AI-specific product roadmaps in upcoming earnings calls. How successfully they navigate the balance between meeting demand and avoiding eventual oversupply will test the thesis that the boom-bust cycle has truly been broken.
Broader adoption of AI across industries could further entrench memory’s new role as a more stable growth sector. However, any significant slowdown in enterprise AI spending or major technological shifts could still test the resilience of the new market dynamics.
The memory industry’s transformation reflects the broader impact of AI on the technology supply chain. What was once a highly cyclical commodity business is increasingly becoming a critical enabler of the AI revolution, with potentially lasting changes to its fundamental economics.